Remove IDNA processing in favor of UTS46

ens/constants.py

@@ -7,6 +7,4 @@
 EMPTY_SHA3_BYTES = b'\0' * 32
 EMPTY_ADDR_HEX = '0x' + '00' * 20
 
-MIN_ETH_LABEL_LENGTH = 7
-
 REVERSE_REGISTRAR_DOMAIN = 'addr.reverse'

ens/utils.py

@@ -14,11 +14,9 @@
     AUCTION_START_GAS_CONSTANT,
     AUCTION_START_GAS_MARGINAL,
     EMPTY_SHA3_BYTES,
-    MIN_ETH_LABEL_LENGTH,
     REVERSE_REGISTRAR_DOMAIN,
 )
 from ens.exceptions import (
-    InvalidLabel,
     InvalidName,
 )
 
@@ -81,10 +79,11 @@ def normalize_name(name):
         return name
     elif isinstance(name, (bytes, bytearray)):
         name = name.decode('utf-8')
+
     try:
-        return idna.decode(name, uts46=True, std3_rules=True)
+        return idna.uts46_remap(name, std3_rules=True)
     except idna.IDNAError as exc:
-        raise InvalidName("%s is an invalid name, because %s" % (name, exc)) from exc
+        raise InvalidName(f"{name} is an invalid name, because {exc}") from exc
 
 
 def is_valid_name(name):
@@ -104,37 +103,6 @@ def is_valid_name(name):
         return False
 
 
-def name_to_label(name, registrar):
-    name = normalize_name(name)
-    if '.' not in name:
-        label = name
-    else:
-        name_pieces = name.split('.')
-        registrar_pieces = registrar.split('.')
-        if len(name_pieces) != len(registrar_pieces) + 1:
-            raise ValueError(
-                "You must specify a label, like 'tickets' "
-                "or a fully-qualified name, like 'tickets.%s'" % registrar
-            )
-        label, *label_registrar = name_pieces
-        if label_registrar != registrar_pieces:
-            raise ValueError("This interface only manages names under .%s " % registrar)
-    return label
-
-
-def dot_eth_label(name):
-    """
-    Convert from a name, like 'ethfinex.eth', to a label, like 'ethfinex'
-    If name is already a label, this should be a noop, except for converting to a string
-    and validating the name syntax.
-    """
-    label = name_to_label(name, registrar='eth')
-    if len(label) < MIN_ETH_LABEL_LENGTH:
-        raise InvalidLabel('name %r is too short' % label)
-    else:
-        return label
-
-
 def to_utc_datetime(timestamp):
     if timestamp:
         return datetime.datetime.fromtimestamp(timestamp, datetime.timezone.utc)

tests/conftest.py

@@ -11,8 +11,6 @@
     get_open_port,
 )
 
-pytest_plugins = ["pytest_ethereum.plugins"]
-
 
 @pytest.fixture(scope="module", params=[lambda x: to_bytes(hexstr=x), identity])
 def address_conversion_func(request):

tests/core/utilities/test_abi_is_encodable.py

@@ -34,9 +34,9 @@
         ('dennisthepeasant.eth', 'address', True),  # passes because eth_utils converts to bytes :/
         ('autonomouscollective.eth', 'address', True),
         ('all-TLDs-valid-now.test', 'address', True),
-        ('-rejects-invalid-names.test', 'address', False),
         ('ff', 'address', True),  # this could theoretically be a top-level domain (TLD)
         ('0xname.eth', 'address', True),  # 0x in name is fine, if it is not a TLD
+        ('rejects_invalid_names.eth', 'address', False),  # no underscore in domain names
 
         # Special bytes<M> behavior
         ('12', 'bytes2', True),  # undersize OK
@@ -66,7 +66,7 @@
         (('1', 0), '(bytes,int128)', False),
 
         (('dennisthepeasant.eth', 0), '(address,int128)', True),
-        (('-rejects-invalid-names.test', 0), '(address,int128)', False),
+        (('rejects_invalid_domains.eth', 0), '(address,int128)', False),
 
         ((b'anything', 0), '(string,int128)', True),
         ((b'\x80', 0), '(string,int128)', False),

tests/ens/test_nameprep.py

@@ -12,12 +12,25 @@ def test_nameprep_basic_unicode(ens):
     assert ens.nameprep("öbb.at") == "öbb.at"
     assert ens.nameprep("Öbb.at") == "öbb.at"
     assert ens.nameprep("O\u0308bb.at") == "öbb.at"
-    assert ens.nameprep("xn--bb-eka.at") == "öbb.at"
     assert ens.nameprep("faß.de") == "faß.de"
     assert ens.nameprep("fass.de") == "fass.de"
-    assert ens.nameprep("xn--fa-hia.de") == "faß.de"
+    assert ens.nameprep("🌈rainbow.eth") == "🌈rainbow.eth"
+    assert ens.nameprep("🐔🐔.tk") == "🐔🐔.tk"
+    assert ens.nameprep("√.com") == "√.com"
+    assert ens.nameprep("ԛәлп.com") == "ԛәлп.com"
+    assert ens.nameprep("test\u200btest.com") == "testtest.com"
+    assert ens.nameprep("-test.com") == "-test.com"
+    assert ens.nameprep("1test.com") == "1test.com"
+    assert ens.nameprep("test.1com") == "test.1com"
 
 
-def test_nameprep_std3_rules(ens):
-    with pytest.raises(InvalidName):
-        ens.nameprep("not=std3")
+@pytest.mark.parametrize(
+    'url', [
+        ('not=std3'),
+        ('not_std3.eth'),  # underscores not allowed
+    ]
+)
+def test_nameprep_std3_rules(ens, url):
+    with pytest.raises(InvalidName,
+                       match=f'{url} is an invalid name'):
+        ens.nameprep(url)